Diet, occupational exposure and early asthma ... - BioMedSearch

Rémen et al. BMC Public Health 2012, 12:387 http://www.biomedcentral.com/1471-2458/12/387

RESEARCH ARTICLE

Open Access

Diet, occupational exposure and early asthma incidence among bakers, pastry makers and hairdressers Thomas Rémen1,2,4*†, Dovi-Stéphanie Acouetey1,2†, Christophe Paris1,2 and Denis Zmirou-Navier1,2,3

Abstract Background: The natural history of occupational asthma (OA) is influenced by many determinants. This study aims to assess the combined roles of personal characteristics, including occupational exposure and nutritional habits, on the incidence of OA during the first years at work. Methods: A nested case–control study was conducted within a retrospective cohort of young workers in the bakery, pastry-making and hairdressing sectors. Cases were subjects diagnosed as ‘confirmed’ or ‘probable’ OA consecutively to a medical visit (N = 31). Controls were subjects without OA (N = 196). Atopy was defined after blood specific IgE analysis, based on the PhadiatopTM test. Occupational exposure was characterized by standardized questionnaires and diet patterns by a food frequency questionnaire. Results: Among bakers and pastry-makers, only atopy is an independent risk factor of OA (OR = 10.07 95%CI [2.76 – 36.65]). Among hairdressers, several variables are associated with OA. Body mass index (unit OR = 1.24 [1.03 – 1.48]) and the score of exposure intensity (unit OR = 1.79 [1.05 – 3.05]) are independent predictors of OA, but the role of atopy is weak (OR = 4.94 [0.66 – 36.75]). Intake of vitamin A is higher among hairdressers cases (crude p = 0.002, adjusted p = 0.01 after control for body mass index and atopy); the same observation is made for vitamin D (crude p = 0.004, adjusted p = 0.01). Conclusion: This study suggests that the influence of several factors on the incidence of OA, including dietary vitamins, might vary across exposure settings. Keywords: Occupational asthma, Epidemiology, Atopy, Vitamins

Background Occupational asthma (OA) is a disease characterized by variable airflow limitation and/or airway hyperresponsiveness due to causes and conditions attributable to a particular work environment rather than to stimuli encountered outside the workplace [1]. Two types of OA can be distinguished: (i) immunologic OA appears after a latency period of exposure necessary for acquiring immunologic sensitization to the causal agent(s); and (ii) non immunologic OA occurs after acute exposure to high concentrations of irritants (‘irritant-induced asthma’) [2]. * Correspondence: [email protected] † Equal contributors 1 Inserm U954 (Institut National de la santé et de la Recherche Médicale), School of Medicine, Nancy, France 2 Lorraine University Medical School, Nancy, France Full list of author information is available at the end of the article

A general framework for the natural history of immunological OA encompasses several stages after onset of exposure, including development of sensitization and inception of OA that can be followed by removal from exposure and remission or persistence of OA [3]. This allergic march is influenced by multiple determinants. While occupational exposure plays the key role, with the interplay between the nature of agents encountered at the workplace (in particular the contrast between high– vs. low molecular weight agents) and intensity or exposure duration before and after occurrence of symptoms, other factors related to personal or more general characters also contribute to the onset of the disease, among which are genetic predispositions and possibly nutritional factors [3]. A large increase in the prevalence of asthma was observed during the last decades in most developed

© 2012 Rémen et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


countries [4]. According to Allan and coll., it would be the consequence of changing environmental and/or lifestyle factors rather than genetic influences [5]. Changes in diet have been put forward as responsible for part of this increase. However, interactions between these factors remain unclear. Numerous hypotheses have been discussed concerning the role of nutrition in asthma occurrence without definitive evidence. The role of antioxidants (vitamins A, C, E) or of vitamin D is still questioned, with contradictory hypotheses being reported [5,6]. Considering poly-unsaturated fatty acids (PUFA), a suggested mechanism relates increased dietary n-6:n-3 PUFA ratio to the augmentation of allergic disease and to asthma [4]. These hypotheses still need to be documented. Most of the epidemiologic studies involving nutritional factors have been implemented in the general population. Very few explored nutritionenvironment interactions in OA [7] which can be viewed as a model of “experimental” asthma [8]. In the framework of a retrospective follow-up of young bakers, pastry-makers and hairdressers, i.e. in occupational sectors with a known high risk of OA, a nested case–control study was undertaken to assess the combined influence of personal characteristics (including nutritional habits) and occupational exposure on the incidence of OA during the first years at work.

Methods Design and study population

The study protocol has been published previously [9]. Briefly, the ABCD (French acronym for early asthma in bakery and hairdressing sectors) study aimed to assess early incidence of OA among young workers according to their sector of activity and exposure duration, and to identify risk factors of OA. It was based on a retrospective follow-up design, with a nested case–control facet. The population based study comprises groups of exposed (bakers, pastry-makers and hairdressers) and non exposed subjects (sales and food sectors: butcher, pork butcher, caterer, cook job. . .), used as a reference group, who graduated between 2001 and 2006 (2001 for the non exposed group) from nine vocational schools in Lorraine, North-Eastern France. The cohort consisted of all who had completed a phone medical questionnaire about respiratory, ENT and skin conditions since engaging in their training and occupation, and about their connection with work [10]. Within this subset of the cohort, all subjects who declared work-related respiratory symptoms or isolated work-related rhinitis symptoms that had appeared after inception of exposure, and a matched sample of all others (frequency matching criteria: year of graduation, vocational school and occupational sector) were invited to participate in a medical visit to complete clinical and

Page 2 of 8

lung function investigations and to collect blood samples for IgE assays (total IgE and specific IgE for work-related and common allergens). The research program is authorized by the Nancy University Hospital ethics committee and written consents are obtained from the young workers themselves. Selection of cases and controls

For the nested case–control study, we included only bakers, pastry-makers and hairdressers who had the medical visit. Diagnosis of OA encompassed two aspects: (i) diagnosis of asthma, and (ii) evidence of a temporal relationship between the occurrence of symptoms and occupation. Each subject who performed the medical visit was classified according to a decisional tree for the definition of OA [10] into one of four categories: (i) confirmed OA; (ii) probable OA; (iii) possible OA; and (iv) absence of OA. Briefly, this classification was established after data collected during the medical visit on the basis of the following criteria: (i) the clinical definition of asthma proposed by the International Primary Care Respiratory Group [11]; (ii) recording of peak-flow expiratory curves for three weeks; (iii) a spirometry with bronchodilatator testing (short acting β-2 agonist); and finally (iv) work-related specific IgE assays when available. Cases were subjects diagnosed as ‘confirmed’ or ‘probable’ OA. Controls were subjects not diagnosed as OA (neither confirmed, probable, nor possible) and nor labelled with ‘unknown’ OA status (when PEF monitoring was not exploitable and reversibility tests were either not feasible or not usable). Data collection

During the medical examination, height and weight were recorded with the same device - electronic balance and electronic height gauge - for all subjects (to avoid measurement bias). Two investigators realized the vast majority of medical visits (TR and DSA), with the assistance of three medical interns for some visits. Data on past and present tobacco smoking, history of work-related respiratory, rhinitis, eczema and urticaria symptoms were collected using the EGEA questionnaire [12]. The time sequence of symptoms with work was assessed using items derived from other dedicated questionnaires [13,14]. Atopic family history (first degree) was considered if at least one of the following pathologies was reported: asthma, hay fever, eczema, urticaria, or in case of a family history of treatment for allergy or desensitisation. From blood sampling, specific IgE analysis was used to categorize subjects as atopic or not, based on the PhadiatopTM test (Phadia, Sweden). This test screens an


IgE-dependant allergy with a median sensitivity of 96% (ranging from 70% to 100% across studies) and a median specificity of 95% (77% to 100%) according to a metaanalysis on a general adult population [15]. The PhadiatopTM test includes mite, pollen, mould, and animal dander allergens, but the exact allergen composition is not issued by the company [15]. The phone screening questionnaire and the medical visit took place from March 2009 to July 2010. Analyses were performed from November 2010 to August 2011. The research program was authorized by the Nancy University Hospital ethics committee and written consents were obtained from the young workers themselves. Exposure assessment

Exposure scores and duration were obtained from the occupational exposure questionnaires [16,17]. A score of exposure intensity depends on the average number of tasks declared to be realized each day. For bakers, the daily tasks were bread-making (kneading-machine loading, transfer by shovel, dough division, dough shaping, bread put in the oven), pastry-making (pastry preparation) and cleaning activities, and were scored (0, 1 or 2) using tertiles [see Additional file 1: Table S1 and Table S2 for more precision]. The sum of these 7 notes (from 0 to 2) defined the score of exposure intensity, a score of 0 reflecting a low exposure, a score of 14 a very high exposure. Calculation of exposure duration entailed reconstruction of the occupational history of each subject since inception of apprenticeship (duration of apprenticeship, diplomas obtained, hiring date, cumulative length of unemployment, date of sector dropout and date of administration of the phone questionnaire). This exposure duration corresponds to the cumulative periods of exposure since engaging in apprenticeship (excluding classroom periods during apprenticeship or periods of inactivity) until the date of the telephone interview. Hence, by construction, one year exposure duration encompasses more than one calendar year. For hairdressers, a corresponding intensity score was based on the following four activities: perm, hair dyes, hair bleached and rinsing, with a note of 0, 1 or 2 according to tertiles of the average number of hairdressing tasks reported per day. The sum of these 4 scores (from 0 to 2) defined the score of exposure intensity among hairdressers, a score of 0 reflecting a low exposure, a score of 8 a very high exposure [see Additional file 1: Table S1 and S2 for more precision]. Nutritional intake

A food frequency questionnaire (Suvimax 2) [18] was used to evaluate dietary imbalance of nutritional factors during the 12 months prior to the medical visit. Briefly, this semi-quantitative food frequency questionnaire

Page 3 of 8

(FFQ) was developed for self-administered assessment of usual dietary intake over the past year among French adults at an individual level. The food list contained 240 food and beverage items categorized into 22 categories. Validity and reproducibility of this questionnaire has been described elsewhere [18]. The frequency of consumption referred to usual consumption over the past year on an increasing scale including yearly, monthly, weekly or daily units, as suitable. The FFQ was selfadministered and completed at home. Questionnaires were recovered and verified during the medical visit in order to complete eventual missing information and correct ambiguities. French recipes, validated by food and nutrition professionals, were used to estimate amounts of simple items consumed among mixed foods. Frequencies were converted into numbers of servings per day and multiplied by the standard portion size or by the portion size declared using photographs. An ad hoc composition table was developed to calculate weighted mean nutrient values. The weights were derived from the gender-specific mean frequency of each item that was declared [18].

Statistical analysis

The data were analysed using SASW 9.2 software. Proportions are expressed as percentages and quantitative data as means with standard deviations (SD). Quantitative data were tested for linearity. For non linear variables, they were divided into categories using quartiles. Comparisons used chi-square or Fisher exact tests, Student or Kruskal-Wallis tests, as appropriate. Logistic regression was used to calculate odds ratios (ORs) and 95% confidence intervals, while adjusting for confounders or exploring effect modification. The significance level to remain in the model was fixed to 0.20, not to overlook confounding and weak associations. Personal characteristics such as gender, age, BMI and smoking habits were compared between cases and controls. Medical characteristics such as atopy, respiratory, ENT and skin symptoms were also compared between these two groups. The role of nutritional intakes was explored in a second step. Crude ORs and, when appropriate, ORs adjusted for potential confounders such as atopy and body mass index are presented. Workrelated symptoms were not retained for the multivariate model because of their strong association with the OA status. Because of differences in the mechanisms involved in the onset of OA between the two occupational sectors, determinants of OA can vary according to the sector. Hence, the analysis was conducted separately according to sector, after a global assessment.


Page 4 of 8

Results Demographics

31 cases met the inclusion criteria. Respiratory symptoms had begun between less than 1 year to 7.4 years prior to the medical visits. Among the 202 control subjects meeting the inclusion criteria, 6 were excluded because of airway responsiveness whose association with occupational nature could not be investigated (absence of specific IgE assays), bringing the total number to 196 controls. Demographic and other relevant characteristics of the 227 subjects (mean age: 25.2 years, 40.5% hairdressers) retained in the analysis are summarised in Table 1. Differences were observed between cases and controls concerning occupational sector and gender (more hairdressers and therefore more women among controls, respectively p = 0.03 and p = 0.02) and sector dropout (higher among cases [p = 0.002]). Among hairdressers retained for the case–control study, 96.7% are women while, among bakers and pastry-makers, 90.4% are men. Because gender and occupational sectors are so strongly associated, no adjustment for sex will be done in sector specific analyses. Atopy, exposure and symptoms

Crude associations are summarised in Table 2, separately for bakers and pastry-makers on the one hand, and hairdressers on the other hand. Atopy is significantly more frequent among cases than among controls in the bakery and pastry-making sectors; the same pattern is seen in hairdressers, but fails to reach significance. Exposure intensity is higher among hairdressing cases. After multivariate analysis, only atopy remains an independent risk factor of OA among bakers and pastrymakers (OR = 10.07 95%CI [2.76 – 36.65]). Among hairdressers, atopy is weakly associated with OA (OR = 4.94

[0.66 – 36.75]), while BMI (unit OR = 1.24 [1.03 – 1.48]) and the score of exposure intensity (unit OR = 1.79 [1.05 – 3.05]) are independent predictors of OA. OA and nutrient intakes

Associations between OA and nutrients intake are presented in Table 3. Overall, the associations of OA with vitamins D and E are borderline significant (p = 0.06 for the two vitamins: 3.8 μg [SD = 3.0] among cases, vs 2.9 μg [2.1] among controls for vitamin D; respectively 21.1 mg [16.7] and 16.3 mg [11.0] for vitamin E), but vanishes after adjustment for body mass index and personal atopy. Adding exposure intensity in the model does not change the effect measures of nutrients intake (data not shown). Among bakers and pastry-makers alone, no nutrient intake is associated with OA, whether prior to or after adjustment. On the other hand, among hairdressers, intake of vitamin A is higher among OA cases before adjustment (p = 0.002; 1.9 mg [1.2] among cases, vs 0.7 μg [0.6] among controls) and after (p = 0.01). The same association is found with vitamin D, whether before (p = 0.004; 5.5 μg [2.5] among cases, vs 2.5 μg [1.7] among controls) or after adjustment (p = 0.01).

Discussion Our results show that among hairdressers, intensity of exposure, measured by the daily number of specific tasks, and body mass index are risk factors of OA. Also, vitamin A and D intakes are greater among cases than among controls in this group. On the other hand, nutritional patterns showed no association with OA among bakers and pastry-makers. While the well known role of atopy on the occurrence of OA is confirmed in this sector, it is also suggested among hairdressers.

Table 1 Descriptive characteristics of cases and controls Characteristic

Cases (N = 31)

Controls (N = 196)

Total (N = 227)

p

Test usedΩ

Women

%

8 (25.8%)

94 (48.0%)

102 (44.9%)

0.02

Chi2

Age (years)

Mean [SD]

26.1 [2.5]

25.1 [3.0]

25.2 [2.9]

0.06

T

Height (cm)

Mean [SD]

173.4 [8.0]

170.1 [9.2]

170.5 [9.1]

0.06

T

Weight (kg)

Mean [SD]

77.5 [15.5]

70.2 [14.1]

71.2 [14.5]

0.009

T

0.38

Chi2

Chi2

Tobacco smoking (at visit): Never smokers

%

10 (32.2%)

76 (38.8%)

86 (37.9%)

Ex smokers

%

7 (22.6%)

26 (13.3%)

33 (14.5%)

Current smokers

%

14 (45.2%)

94 (48.0%)

108 (47.6%)

Hairdressers

%

7 (22.6%)

85 (43.4%)

92 (40.5%)

0.03

Exposure duration (years)

Mean [SD]

6.1 [3.1]

6.7 [2.2]

6.6 [2.4]

0.25

KW

Sector dropout†

%

10 (32.3%)

22 (11.2%)

32 (14.1%)

0.004

F

Occupation:

Ω Chi2 = chi-square test – T = Student test – KW = Kruskal-Wallis test – Fisher = fisher test. † From the sector in which the subjects graduated.


Page 5 of 8

Table 2 Distribution of symptoms and of putative risk factors among cases and controls according to the occupational sector Variable

All Cases (N = 31)

Controls (N = 196)

p

Hairdressers

Controls (N = 111)

p

Cases (N = 7)

Controls (N = 85)

p

Test usedΩ

25.6 [5.8]

25.1 [4.1]

0.94

27.2 [6.3]

23.1 [3.4]

0.07

KW

Body mass index (kg/m2)

Mean [SD] 25.9 [5.9]

24.2 [3.9]

Work-related rhinitis symptoms

%

28 (90.3%)

77 (39.3%)